Find the pages or queries behind a change
Choose two Pages or two Queries exports. Your files stay in this browser tab.
Earlier period
Or paste CSV
Recent period
Or paste CSV
English headers · up to 5,000 rows and 2 MB per file · separate, single-period exports. Comparisons use clicks per day.
Prepare two comparable exports
- Open Search Console’s Search results report. Choose the property, search type, country, device, and other filters you want to compare.
- Select an earlier, complete period with date comparison off. Export the report as CSV. Extract the ZIP if needed and keep
Pages.csvorQueries.csv. - Change only the dates to your recent, complete period and export again. Use the same table in both files.
- Enter each period’s day count, including both endpoints. Two four-week periods make a useful starting comparison because they contain the same number of each weekday. For seasonal businesses, also review a comparable period last year.
This tool reads English headers: Page or Query, Clicks, Impressions, and Position. CTR is recalculated. It also accepts tab-separated spreadsheet cells and semicolon-separated files. Each file can contain up to 5,000 unique rows and 2 MB. Combined comparison exports, date tables, and query-plus-page tables are not supported.
Google’s export documentation explains its export limits. Accepting a larger file here cannot recover rows Google omitted. For format examples, use “Try an example” and open the paste fields.
A 350-click decline can contain 450 lost clicks
Here is the fictional example built into the tool. Both periods last 30 days, and four pages appear in both files.
| Page | Earlier clicks | Recent clicks | Change |
|---|---|---|---|
| Grinder guide | 600 | 300 | −300 |
| Cleaning guide | 300 | 150 | −150 |
| Coffee shop | 200 | 300 | +100 |
| About | 100 | 100 | 0 |
| Matched total | 1,200 | 850 | −350 |
The two guides lost 450 clicks, while the shop gained 100. The net change is −350. The grinder guide accounts for two thirds of the loss among declining pages—not two thirds of the net change. Mixing those denominators produces confusing percentages.
The grinder guide’s impressions stay at 20,000, but CTR falls from 3% to 1.5%. The cleaning guide’s CTR stays at 3%, while impressions halve. Those pages deserve different investigations even though both lost clicks.
An old sale page appears only in the earlier export, and a new guide appears only in the recent one. Neither gets an invented zero. Their observed numbers remain available under “Only in one export” and in the download.
Use the pattern to choose the next check
| Observed pattern | Useful next check |
|---|---|
| Fewer impressions | Compare rankings and the mix of queries. Check seasonal demand, then inspect important URLs if visibility disappeared. |
| Lower CTR | Compare the same queries and devices. Review what the search result promises, competing results, and changes in search appearance. |
| Both fell | Start with the largest declining page, then narrow to its queries. Several things may have changed together. |
| Average position worsened | Check individual queries before concluding that every ranking fell. New low-ranking queries can change the average. |
| Only in one export | Search for the exact row in the other period inside Search Console. Check filters, row limits, and any URL changes. |
The labels describe the numbers; they do not prove a penalty, an algorithm update, lost search demand, or an AI Overview effect. Google’s traffic-drop investigation guide includes wider checks for technical problems, reporting anomalies, and seasonality.
If clicks fell while impressions rose, our worked query-mix example is a useful next step. If important pages have disappeared, use the indexing checklist.
How the comparison works
Daily click change = recent clicks ÷ recent days − earlier clicks ÷ earlier days
URLs match exactly, including their path, parameters, and trailing slash. Queries match without regard to case. Both files must have one row per item. Duplicate labels fail explicitly. Renamed URLs remain unmatched; the tool does not assume two different addresses are the same page.
Clicks and impressions are compared per day. CTR is clicks divided by impressions within each period; its difference is shown in percentage points. Position is the exported average, and a lower number is better. Zero-impression rows have no CTR or meaningful position comparison.
For example, 300 clicks over 30 days and 150 over 15 days are both 10 clicks per day. That removes the period-length mismatch, but it does not remove seasonality or differences between weekdays. Choose sensible periods before interpreting the result.
Summary totals use only matched rows. The page-section view groups those rows by host and first path segment. Neither view is a whole-site traffic total. Read why Search Console totals can differ from exported rows if the numbers do not match the chart.
Your files are read in this browser tab. They are not uploaded, saved to browser storage, or sent to an AI service. Clear or reload the page to discard them. Downloads include every comparison row, period lengths, and unmatched labels, regardless of the current table filter.